Following an investigation into research misconduct, the Journal of Clinical Investigation has retracted a cancer genetics paper from a laboratory at the National Institutes of Health due to “data falsification and fabrication” of four figures and a table in the paper.
The paper, “FOXO3 programs tumor-associated DCs to become tolerogenic in human and murine prostate cancer,” describes an overexpressed gene in mouse prostate cancers that appears to suppress immune system cells.
The journal retracted the paper following an investigation into author Stephanie K. Watkins, then a postdoctoral fellow at the National Cancer Institute. According to a NIH press release released about the study in March 2011, the work “has led to the submission of a patent application by the NIH on behalf of Hurwitz and Watkins to target FOXO3 as a way to boost immune responses in cancer and to silence excessive immune responses in autoimmune diseases.” We found an NIH record of the patent application, but no record of an approved patent at the United States Patent and Trademark Office under either Hurwitz or Watkins’ names.
The paper has been cited 62 times, according to Thomson Scientific’s Web of Knowledge. Here’s the full notice:
At the request of the corresponding author, the JCI is retracting the paper “FOXO3 programs tumor-associated DCs to become tolerogenic in human and murine prostate cancer” based on findings of data falsification and fabrication regarding Figures 1D and 8A, likely data falsification regarding Figure 4A, likely falsification or fabrication regarding Figure 4B, and erroneous sample attribution in Table 1. Following extensive review by an NIH-appointed investigation committee, the NIH found that one author, Stephanie Watkins, was the sole individual responsible for the instances of research misconduct. None of the other authors was aware of the misconduct.
We could not find contact information for Dr. Watkins, though she now appears to be an assistant professor at Loyola University in Chicago, with the retracted paper listed under her publications. Corresponding author Arthur Hurwitz directed us to Dr. Melissa Colbert, the NIH Research Integrity Officer, who had only this to say:
NIH believes that the retraction notice speaks for itself and has no further comment. Research misconduct proceedings at the NIH are considered confidential.
We also reached out to the JCI and will update if we hear back from them.
HA!!! The uber-hype press release from the NIH in 2011 has also been retracted. HILARIOUS!
“Uh… nevermind.”
I have repeatedly questioned the responsibility of the lead author, PI and the leader of the laboratory, whyonly one person is singled out. This is the prime responsibikity of the leader of the team to check and confirm the validity of data. Moreover, while submitting a manuscript for every journal, exah author has to sign a form that also applies to JCI where this article was published. As noted below, the other autrhors signed that they have verified the data, this aspect should be discussed.
Authorship responsibility. I certify that I have participated sufficiently
in the conception and design of this work and the analysis of the data
(where applicable and within my area of expertise), or the writing of the
manuscript, and to take public responsibility for it. I warrant that my work
is original and scientifically accurate, that it is not defamatory, does not
invade any right of privacy, and does not infringe on any proprietary right
or copyright. I agree to indemnify and hold the ASCI and the JCI harmless
against any claim to the contrary. I believe the manuscript represents valid
work. I have reviewed the final version of the manuscript and approve it
for publication. To my knowledge, neither this manuscript nor one with
substantially similar content under my authorship has been published
or is being considered for publication elsewhere, except as described in
an attached letter. I attest that I shall produce the data on which the
manuscript is based for examination by the editors or their assignees
should they request it.
It is fine to argue that the PI apparently was not suspicious enough of his postdoc when he should have been, but responsibility in my opinion requires something more than that. Especially when the person who committed the fraud was a postdoc, not some inexperienced trainee scientist.
But the ultimate responsibility is of the lead author too check and confirm the data because ultimately it is his research and psotdoc is assisting him to accomplish his goals.
It is as much the research of the first author as that of the PI. Again, this is a postdoc, someone who is expected to take responsibility for what they do. They are supposed to show in that phase that they can do research independently.
There is also a limit to how much data checking is possible. Ultimately you have to trust someone somewhere in the process. If my postdoc writes in his labbook he weighed 5.03 mg of compound A, I have no way of checking that. I could then do all experiments myself all over again, but when would that ever happen? Never, and thus the demands the PI checks all data – which in essence comes down to repeating all experiments themselves, personally – would mean the poor postdoc would never see a publication come out of his/her work, because the PI does not have the time to check all the results.
It’s not hard to fool a senior author with falsified data. These mouse tumor experiments take months, how would you check them? Hire a second postdoc who is willing to waste months/years of their life repeating someone else’s work? And when you do, ask them how they expect to get another science job when they’ve spent so long repeating someone else’s work and have no first author papers. This is the real world.
Ah, but will you believe the results of postdoc #2? Surely you will then have to hire a postdoc #3, 4, 5, 6, ad infinitum to check the results of the previous one? Who knows which one of them cheated, if they even did?
Good point. Perhaps only finalise each figure when three postdocs, who have never met, achieve exactly the same result. And when you eventually have 500 authors, make them all first author. Job done.
I hope to live to read one day a retraction notice stating something outrageous and improbable of a kind:
“the investigation committee found that the inappropriate reward and punishment system ran by the PI as well as the unhealthy climate of excessive competitiveness in the lab makes it difficult to determine the sole individual responsible for the instances of research misconduct. It also to be assumed that at least some of the other authors neglected their duties by purposefully looking the other way despite having reasons to suspect or even being aware of the misconduct”.
I’m curious as to how far you would take this principle? If a postdoc manipulates a figure, do you think all co-authors should be assumed to have “neglected their duties by purposefully looking the other way “? If you were the postdoc who had done that, do you think the other co-authors should bear some of the responsibility?
Don’t forget that you are commenting on a real case, with real people. The NIH notice makes it clear that Stephanie Watkins was the lone individual responsible for falsifying most of a paper. And as some figures are listed as “likely” false, she can’t be co-operating with the committee even now or they would at least know what was really faked. Her co-authors have most likely suffered a great deal. I don’t believe it is fair to rub salt in the wound and imply (without any basis) that the PI is to blame for fostering an unhealthy environment and that the co-authors looked the other way.
This thread reveals a key issue. The ORI insists on confidentiality to a point of absurdity. This notice bluntly informs us that Stephanie Watkins was alone responsible for faking this quite high impact paper, so the confidentiality isn’t going to do her much good now. The internet (and is presumably a reflection of the wider world) then arbitrarily points the finger at the PI and co-authors, who have done nothing wrong according to the only information we have, so the confidentiality isn’t doing them any good. Perhaps the confidentiality only benefits the NIH itself? Certainly not the scientific community.
Shining some real light into these situations would be very useful. As some have said, sometimes unhealthy competitive environments can force postdocs into it. But sometimes postdocs themselves can foster an environment of fear and intimidation (if you’ve never come across this, ask around, you won’t have to look far). Sometimes the co-authors are genuine victims.
Why not release full details of the investigation once it has concluded? It would be helpful to us all.
That and the investigations take so long to complete that people can go off and get another job (or two) by the time their investigations are resolved. Meanwhile, anyone aware of the ORI investigation has to keep quiet. It would be better not to do a federal investigation if all it does is prevent someone’s former department from making a potential employer aware that they were dismissed for fraud.
WayShe, the fact is that ORI has not “conducted investigations” to any significant degree since the 1999 decision of HHS officials to give the authority to each agency for its intramural programs, and to the HHS OIG for extramural programs, to conduct any necessary investigative fact-finding for HHS [ see my paper on ORI history in Accountability in Research 20, 291-319 [available online at: http://www.tandfonline.com]. ORI conducts aggressive oversight review of those investigations by HHS agencies and outside research institutions, before deciding whether to make ORI/HHS findings of research misconduct and to impose administrative actions on the person who committed the falsification, fabrication, or plagiarism.
Furthermore, WayShe, ORI does not “do a federal investigation .. . [that would] prevent someone’s former department from making a potential employer aware that they were dismissed for fraud.” It is entirely the authority and decision of the institution that made is finding of research misconduct as to who has a need to be informed of their decision. ORI only demands that its process remain confidential, until and if ORI makes its own finding.
Nm, I was talking in general, since such kind of notices, i.e., placing blame exclusively on a single individual low in hierarchy, is rather common.
But I’ve spent enough time in a lab as well to know what I am talking about. The rogue PhD or postdoc must be very sinister, sneaky and cunning indeed to fool own PI (who is surprisingly often a controlling type), without ever raising suspicion. Again, such case may or may not have happened here. You state correctly, we cannot know what really happened, due to confidentiality (indeed, for whom and what exactly?)
Nm, it is not true that “the ORI insists on confidentiality to a point of absurdity” — ORI is the only federal agency [NSF OIG and the others do not] that publishes its findings of research misconduct, naming the person who falsified or fabricated or plagiarized the research, and describing exactly the results involving research misconduct, and the federal administrative action imposed by ORI/HHS (often debarment from federal funding). This an ORI policy that I implemented for ORI beginning in 1992, publishing our findings in the Federal Register, NIH Guide to Grants and Contracts, ORI Newsletter and ORI Annual Report, and the online PHS Administrative Actions Bulletin Board — all of which can be found through the ORI website (http://ori.hhs.gov).
However, ORI is required to follow the HHS regulations [http://ori.hhs.gov/FR_Doc_05-9643] that demand confidentiality (to protect the identity of the complainant and the respondent/accused, who is presumed to be innocent) — until the respondent is proven by ORI to have committed misconduct, after an institutional investigation finding misconduct). Then ORI makes public its findings.
Nonetheless, the institution is not restricted by ORI from making public statements about its own findings of misconduct. Obviously this case is one — scientists at NIH (the institution) apparently informed the journal editor of the completed NIH investigation which found one author had committed research misconduct, and that information was included in the retraction notice.
I think you miss the problem. Yes, NIH does publish the results of its investigations. However, should it really be taking 5+ years for some of them to conclude?
No, WayShe, NIH seldom “publishes the results of its investigations” — but ORI publishes all its findings of research misconduct, based on and extending NIH investigations or other institutions’ investigations. Yes, it is unfortunate that some NIH and institutional investigations take several years to complete, and then ORI’s oversight review and pursuit of ORI findings may take several more years in complex cases [often those with senior respondents (and their attorneys) who contest the proposed ORI findings and even appeal them to HHS Administrative Law Judges for potential hearings over many years]. Being thorough and fair is difficult for ORI and other federal agencies, as well as institutions.
The administrative problems at ORI are now widely known along with their impact. It’s not simply a matter of those under investigation lawyering up or that everyone is just trying to do a fair job. When you have irrefutable evidence sitting on your desk for several years, you are doing a disservice to science.
Regarding charging and appeal at HHS, these are actual crimes in many cases. When someone is charged with murder, it’s on the public record. If someone is under investigation by ORI, this should be a matter of public record. To take a different sort of example, suppose you hired a nanny only to find out later that s/he was being prosecuted for child endangerment. The investigation drags on for years in secret. Meanwhile, the nanny is shaking your baby. You catch the nanny doing this and fire him/her, but the nanny just moves on to another family and is shaking their baby. In the real world, you could actually call this new family and warn them. With ORI investigations, those who know the truth are forced to sit idly by while criminals continue in their shenanigans.
Well, WayShe, you can go ahead and make your statements to the public (press, PubPeer, Retraction Watch, SciFraud, etc. while ORI does its hard work and tries to be fair and meet the legal requirements for making or pursuing its findings of research misconduct. ORI cannot stop you from doing this, even though it may not be consistent with the regulations for institutions and their employees. You just have to be willing to live the the consequences yourself when you go public.
The press release is still available in the wayback machine
http://web.archive.org/web/20110727121324/http://www.nih.gov/news/health/mar2011/nci-23.htm
Most sad indeed that such hype happens in the NIH. Fewer studies, performed more methodically without “publish or perish” mentality, would certainly yield a better rate of return for tax dollars spent. When people push to run government groups “like a business” this kind of outcome is the result.
That the NIH is even putting out such hyped press releases is cause enough for concern.
From the press release:
“NCI leads the National Cancer Program and the NIH effort to dramatically reduce the burden of cancer and improve the lives of cancer patients and their families, through research into prevention and cancer biology, the development of new interventions, and the training and mentoring of new researchers.”
All this study did was dramatically reduce some grant budgets, and pad the bottom line of a for-profit journal.
Steve wrote “All this study did was dramatically reduce some grant budgets….”
This is incorrect, and based upon a common misconception. Intramural funding at NIH is fixed by Congress (15% of total, I believe). There is no competition against grant funding.
I retract the word “grant” from my comment.
After the scandal with the Anil Potti gentleman, which was where I came in, to all the other stories of the last five years, we come to the NIH having to retract a paper because a postdoc apparently fabricated data in (some of) the figures: “I’m shocked, shocked, to find that gambling is going on here.”
Retraction Watch, five years ago a curiosity, has snowballed into a regular item, no longer news but more like the weather report.
Am I being cynical, is it just the ascertainment bias, or are things getting worse?
Look at how they actually catch the majority of these cases. Poorly photoshopped Westerns? Duplicated figures/images? Plagiarised graphs? Clearly only the amateur fakers are caught. Anyone with an ounce of sense would just misload their Western, make a new image and mislabel it, or falsify a graph from scratch. The most remarkable thing about pubpeer is not that fraud exists, it is that anyone bothering to fake their data would leave evidence in the paper itself!
My point is, we can’t tell if things are getting worse or not as we can’t possibly be catching more than the tip of the fraud iceberg. Which is depressing.
Reply to “Sadness” above:
I don’t think the Photoshop issues are necessarily sign of amateurism. Misloaded western- this can only be done by someone in the lab, aka “rogue postdoc”. This surely happens often enough. But, what do you do if you are PI, stuck in your office and in need for a certain result/figure? Involve conspirators among your employees to misload a western and pray they don’t sing instead? No, you take the data you were brought and make the best with it. Sometimes you nudge it a bit with Photoshop. If it ever comes out years later, your people will hardly complain if such papers helped their careers as well.
I agree.
Rogue PI would probably have to photoshop. Most rogue postdocs would presumably misload.
The real source of our discussion seems to be which is more likely in general: rogue PI or rogue postdoc. As we don’t have any stats, I guess our prejudices are merely personal experience. But we all presumably agree that both exist.
The question is: why don’t we believe notices such as this one which is worded as unambiguously as it could be? Do RW commentators not trust these investigations? Why not? Is it because, as Leonid has said, the rogue postdoc would have to be “very sinister, sneaky and cunning” and we don’t want to believe that of the woman named here? I mean look at the quote: “the NIH found that one author, Stephanie Watkins, was the sole individual responsible for the instances of research misconduct. None of the other authors was aware of the misconduct.”
On an unrelated note, shouldn’t that be “were aware”? Or do I need to brush up my grammar?
A lot of comments here implying the P.I. got off scot-free; however, if you look closely at the NCI website, it looks like he’s no longer there:
https://ccr.cancer.gov/arthur-andrew-hurwitz
Reply to “DefendSmallScience” above:
Good point, although it doesn’t imply blame. The PI’s career is going to take a huge hit from a “rogue postdoc” whatever the situation (unless the PI is an influential “big name”). Interesting that, as far as RW reports, Stephanie Watkins still has a job as a professor at Loyola (I might skip reading the papers that come out of that) despite being reported as solely to blame… But others seem to be suffering fallout. Life is cruel.
Continuing discussion with “Sadness”:
my personal view is: cases of a sole responsible rogue scientist are probably rare. It is the current competitive and elitist biomedical system which feeds on sensationalism. This system drives and selects for success at any cost and silently breeds and condones misconduct. Those are not even conspiracies, because the only truly betrayed one is the science itself. Everyone else profits (in the short term): scientists get their careers, journals their hot papers, politicians their “success” stories of excellence funding, public gets to read of amazing therapy developments….
Back to the original topic: this is why I am not entirely convinced by that retraction notice.
Just as a quick followup to the article: if you search the patent office’s PAIR system (http://portal.uspto.gov/pair/PublicPair) by the application number (13/520904) provided from the NIH link, you’ll find that the patent application was allowed, but the NIH abandoned the application on November 2014 by not paying the allowance fees needed to get it issued.
In an odd twist, this retraction itself has now been cited by “Regulatory Dendritic Cells in Autoimmunity: A comprehensive review. Liu and Cao”. But the reference appears to view the retracted work positively. Presumably the authors of this “comprehensive review” only read the retraction’s title and so assumed that the retraction was an original research article.
So there we have it. Have your papers retracted, no one will read more than the title and you can get some new citations from the old fraud.
Two 2016 retractions shared authors.
https://pubpeer.com/publications/4D8D828FF37CC6D8DE6AD8BB9FF17D#fb48735
https://pubpeer.com/publications/79D9F8986288DAA03296A4FCE07689#fb48734